A numerical scheme is presented for imaging the three-dimensional distribution of gamma-ray activity under a section of a civil structure, to determine the degree of contamination, and the effectiveness of consequent decontamination. Imaging is to be accomplished by monitoring radiation emission alongside the exposed contaminated surface. In this problem, the number of independent measurements is limited, and can lead to an incomplete imaging problem, in which the expected contribution of image voxels exponentially decreases with depth. A maximum-likelihood expectation-maximization (MLEM) iterative technique was adopted for image reconstruction, guided by supplementary information and aided with a measurement scanning strategy devised to reduce measurement redundancy. The resulting inverse problem was solved for a number of configurations to demonstrate that the algorithm can tolerate measurement noise, nonuniform and sparse contamination conditions, and a low-energy emitting isotope.